Sony AI Table Tennis Robot Ace Takes On Elite Players, Hinting at What Physical AI Can Do Next

Engineers at Sony AI have built a table tennis robot called Ace that can compete with highly trained human opponents, marking a notable step forward for physical AI in fast, real-world settings.

In tests, the system won three of five matches against elite amateur players who train heavily each week.

The results, detailed in a peer-reviewed paper in Nature, reflect progress beyond game-playing software and into robotics that must sense, decide and move under tight time limits.

Table tennis is a particularly demanding benchmark because the ball’s speed, spin and bounce can change from one stroke to the next.

A robot that reads spin in real time

Ace combines high-speed perception with an autonomous decision system and a multi-jointed robotic arm to execute returns with precision.

Researchers highlight its ability to detect and respond to spin, a factor that has limited many earlier table tennis robots in realistic play.

Rather than relying on fixed, pre-programmed shot patterns, Ace uses reinforcement learning to choose actions based on the unfolding rally. That approach allows the robot to adapt to different styles and recover when a point develops in unexpected ways.

How Ace performed against humans?

Sony AI evaluated Ace in matches against seven players, including five elite amateurs and two professionals from Japan. Across 13 games against the elite group, the robot won seven games and converted those into three match wins.

Against the two professional players, Ace was less successful, winning one of seven games and losing both matches overall. Still, researchers say the gap between competent human play and robotic performance is narrowing, especially in handling difficult spinning shots.

Why the breakthrough matters beyond sport?

Project leaders argue the significance is not table tennis itself, but the demonstration of perception and control in a dynamic physical environment.

Systems that can react at speed while maintaining accuracy could translate to tasks such as rapid sorting, precision handling, and safer human-robot interaction in shared spaces.

For now, Ace remains a research platform rather than a commercial product, and it has not reached the level implied by top professional competition.

But by proving it can challenge elite players under real conditions, Sony AI has provided a clearer view of what the next generation of physical AI may look like.